TL;DR
Get tech for your team delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
Some of Anthropic’s earliest employees have reportedly discussed buying remote land in the United States as a place to relocate if AI goes awry. The reported plans are under consideration; the available account does not establish that anyone has bought land or set a move date.
Some of Anthropic’s earliest employees have discussed buying land in remote parts of the United States where they could relocate if AI goes awry, according to a Wall Street Journal report summarized by The Decoder. The plans are described as being considered, not completed, and offer a concrete example of how some people close to frontier AI research think about potential catastrophic risks.
The report says some of Anthropic’s earliest employees told an industry colleague in recent weeks that they were considering remote U.S. land as a possible place to move. The stated contingency was that AI might “go awry.” The account does not name the employees, identify a location, or say that a purchase has been made.
The idea of relocating in response to AI risks reportedly predates this land discussion. Former employees recalled that, at dinners during Anthropic’s early period, people discussed a possible Manhattan Project-style scenario: researchers could be asked to move to a desert location and continue developing AI inside an electromagnetically shielded government facility. That recollection describes a scenario discussed, not an actual government plan or facility.
The reporting connects some early Anthropic and OpenAI hires with Effective Altruism and a Bay Area circle focused on catastrophic risks. It describes people in that circle as having considered threats ranging from asteroid impacts and supervolcanoes to uncontrolled AI. The account does not establish that all early employees share these views or are making similar preparations.
How AI Risk Shapes Personal Plans
The report matters because it describes individuals associated with frontier AI development treating severe AI risk as a scenario worth planning around. Considering a possible place to relocate is a personal contingency measure; by itself, it does not show that the employees have evidence of an imminent crisis or that the company has adopted a relocation policy.
For readers following AI safety, the account also illustrates how concerns about advanced AI can extend beyond research and policy debates into private decisions. The reported plans may prompt questions about how seriously people in the industry assess low-probability, high-impact risks. The available material does not provide enough detail to judge the employees’ reasoning, the level of urgency they assign to the risk, or whether the land idea is practical.
From Catastrophic Risks to AI
The report places the land discussions within a network shaped by Effective Altruism, a movement whose members have often focused on reducing large-scale risks. It says some of Anthropic’s and OpenAI’s earliest hires were connected to that culture, including people in a Bay Area group described as “doomers.” That label is the report’s characterization of the group; it should not be read as a description of every employee at either company.
The account traces some of the group’s concern about uncontrolled AI to Eliezer Yudkowsky, who began warning about the possibility in the mid-2000s. Before AI became a central focus, members of this risk-oriented community also considered natural catastrophes such as asteroid impacts and supervolcanoes. This history helps explain the reported interest in contingencies, but does not confirm that a specific threat has materialized.
Purchase Plans and Risk Assessment
It is not clear whether any employee has bought land, how many people are considering it, where they might relocate, or what timeline they have in mind. The report as summarized does not give the employees’ names or include direct comments from Anthropic about the discussions.
The basis for the reported contingency also remains unspecified. There is no information here about a particular AI system, forecast, threshold, or event that would trigger a move. The reported conversations and past dinner discussions do not establish that AI is currently out of control or that the company expects an imminent emergency.
Whether Discussion Becomes Action
The next development to watch is whether those considering the idea make a purchase or take other practical steps. The available report gives no purchase date, relocation schedule, or stated milestone, so the plans’ status may remain uncertain unless the employees or Anthropic provide further information.
Any additional reporting could clarify how many people are involved, what they mean by AI “going awry,” and whether the land idea is a serious contingency or an exploratory conversation. Until then, the confirmed news is limited to a report of consideration and earlier discussions, rather than a completed relocation plan.
Key Questions
What are some Anthropic veterans reportedly considering?
Some of Anthropic’s earliest employees have reportedly discussed buying remote U.S. land as a place to relocate if AI goes awry, according to the Wall Street Journal report summarized by The Decoder.
Have they bought the land?
The available report says they are considering a purchase. It does not confirm that anyone has bought land.
What does “AI goes awry” mean in this report?
The account does not define a specific event, system, or risk threshold. It presents the phrase as the employees’ reported reason for considering a contingency location.
Was a government facility part of the reported plan?
Former employees reportedly recalled dinner discussions about a hypothetical Manhattan Project-style effort in a shielded government facility. The account does not say that such a facility exists or that the government has made such a plan.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
